The differences as per the last patch: linux32: https://public-artifacts.taskcluster.net/QXHiZbM5TTyPL_-W7c2LuA/0/public/diff.html linux64: https://public-artifacts.taskcluster.net/TT0jN2hrSqyx61_kp3lKqw/0/public/diff.html Note that contrary to bug 1427316, where we were comparing two builds on the same try push, one with the modified GCC and one with one from a try from the parent changeset, here we are comparing against the try for bug 1427316. As a consequence, there are differences related to MOZ_BUILD_DATE and the try changeset being different. So, for an explanation of the differences: - Differences in Build ID in notes sections and .gnu_debuglink are again due to the differences in the other sections of the binaries. - On x86, all the other differences in executables and libraries are due to MOZ_BUILD_DATE - CRC differences in omni.ja are because of the file content differences further below. - omni.ja:modules/AppConstants.jsm, omni.ja:chrome/toolkit/content/global/buildconfig.html only vary because of the try changeset difference. - application.ini and platform.ini vary because of both MOZ_BUILD_DATE and the try changeset difference. - on x86-64, the remaining differences in executables and libraries are to symbol names that are generated by the compiler for thunks, and only exposed because we don't strip them because of --enable-profiling. Here's another diff on linux64 on stripped binaries, which look more like the linux32 one: https://public-artifacts.taskcluster.net/JuBosU9QQu6QKWou0UkHOw/0/public/diff.html I'm not sure why more binaries have differences in .gnu_debuglink, but it doesn't really matter in practice.

Comment on attachment 8939075 [details] Bug 1427339 - Configure binutils and gcc --with-sysroot=/. https://reviewboard.mozilla.org/r/209498/#review215218

Comment on attachment 8939076 [details] Bug 1427339 - Make mozconfig.stdcxx work with both CentOS and Debian-built GCCs. https://reviewboard.mozilla.org/r/209500/#review215220

Comment on attachment 8939077 [details] Bug 1427339 - Build GCC on a Debian-based docker image. https://reviewboard.mozilla.org/r/209502/#review215222

Comment on attachment 8939075 [details] Bug 1427339 - Configure binutils and gcc --with-sysroot=/. https://reviewboard.mozilla.org/r/209498/#review215238 ::: commit-message-677a4:7 (Diff revision 1) > +doesn't work with a binutils built without. However, a GCC built without > +--with-sysroot=/ works fine with a binutils built with it. So this I might have misspoken here. This may be breaking e.g. rusttests builds. Which might actually be a good thing. It's an indicator they're not using the binutils they're supposed to use.

Pushed by mh@glandium.org: https://hg.mozilla.org/integration/mozilla-inbound/rev/f1a65f4c4d3d Configure binutils and gcc --with-sysroot=/. r=gps https://hg.mozilla.org/integration/mozilla-inbound/rev/d20521b7b865 Make mozconfig.stdcxx work with both CentOS and Debian-built GCCs. r=gps https://hg.mozilla.org/integration/mozilla-inbound/rev/0894fbaddc88 Build GCC on a Debian-based docker image. r=gps

https://hg.mozilla.org/mozilla-central/rev/f1a65f4c4d3d https://hg.mozilla.org/mozilla-central/rev/d20521b7b865 https://hg.mozilla.org/mozilla-central/rev/0894fbaddc88

Pushed by mozilla@jorgk.com: https://hg.mozilla.org/comm-central/rev/50504ece5f31 Keep build files in sync (Port 1427339: Make mozconfig.stdcxx work with both CentOS and Debian-built GCCs). rs=bustage-fix